How karankhatavkar/predictive-maintenance-bearing-tool-wear works
This repository is a notebook-based predictive-maintenance study, not a deployed application. Its two documented workflows analyze NASA IMS bearing degradation and PHM 2010 CNC tool wear, then compare traditional and deep-learning predictions across held-out hardware. The graph reflects capabilities and data relationships documented in the README; notebook internals were not available, so it does not infer code-level calls or transfers between notebooks.
Main entry points: 03_deep_learning_tcn_xai.ipynb Deep architectures · 02_traditional_ml.ipynb Vibration features · 03_deep_learning_cnn_lstm.ipynb Raw-signal models · 02_feature_engineering_ml.ipynb Cut features
Components 16
NASA IMS Bearings
- Degradation EDA
nasa-ims/01_eda.ipynb - Vibration features
nasa-ims/02_traditional_ml.ipynb - Traditional ML
nasa-ims/02_traditional_ml.ipynb - Held-out bearing test
nasa-ims/02_traditional_ml.ipynb - Raw-signal models
nasa-ims/03_deep_learning_cnn_lstm.ipynb - Held-out set test
nasa-ims/03_deep_learning_cnn_lstm.ipynb
PHM Tool Wear
- Signal EDA
phm2010/01_eda.ipynb - Cut features
phm2010/02_feature_engineering_ml.ipynb - Traditional ML
phm2010/02_feature_engineering_ml.ipynb - Held-out cutter test
phm2010/02_feature_engineering_ml.ipynb - Deep architectures
phm2010/03_deep_learning_tcn_xai.ipynb - Saliency analysis
phm2010/03_deep_learning_tcn_xai.ipynb - Held-out cutter test
phm2010/03_deep_learning_tcn_xai.ipynb
Other components
- Analyst
- NASA IMS data
- PHM 2010 data
Connections 19
- Analyst → Degradation EDA: analyzes
- Analyst → Vibration features: models
- Analyst → Raw-signal models: models
- NASA IMS data → Degradation EDA: raw signals
- NASA IMS data → Vibration features: bearing signals
- Vibration features → Traditional ML: extracts features
- NASA IMS data → Raw-signal models: raw signals
- Traditional ML → Held-out bearing test: compared on
- Raw-signal models → Held-out set test: tested on
- Analyst → Signal EDA: analyzes
- Analyst → Cut features: models
- Analyst → Deep architectures: models
- PHM 2010 data → Signal EDA: raw signals
- PHM 2010 data → Cut features: cut signals
- Cut features → Traditional ML: extracts features
- PHM 2010 data → Deep architectures: signal data
- Traditional ML → Held-out cutter test: compared on
- Deep architectures → Held-out cutter test: tested on
- Deep architectures → Saliency analysis: explained by